structure

app.py

@@ -1,132 +1,178 @@
-# import gradio as gr
+import gradio as gr
+import os
 # from transformers import AutoTokenizer
-
-# # 第一个功能：基于输入文本和对应的损失值对文本进行着色展示
-# def color_text(text_list=["hi", "FreshEval"], loss_list=[0.1,0.7]):
-#     """
-#     根据损失值为文本着色。
-#     """
-#     highlighted_text = []
-#     for text, loss in zip(text_list, loss_list):
-#         # color = "#FF0000" if float(loss) > 0.5 else "#00FF00"
-#         color=loss
-#         highlighted_text.append({"text": text, "bg_color": color})
-#     return gr.HighlightedText(highlighted_text).get_html()
-
-# # 第二个功能：根据 ID 列表和 tokenizer 将 ID 转换为文本，并展示
-# def get_text(ids_list=[0.1,0.7], tokenizer=None):
-#     """
-#     给定一个 ID 列表和 tokenizer 名称，将这些 ID 转换成文本。
-#     """
-#     return ['Hi', 'Adam']
-#     # tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
-#     # text = tokenizer.decode(eval(ids_list), skip_special_tokens=True)
-#     # 这里只是简单地返回文本，但是可以根据实际需求添加颜色或其他样式
-#     # return text
-
-
-# def get_ids_loss(text, tokenizer, model):
-#     """
-#     给定一个文本，返回其对应的 IDs 和损失值。
-#     """
-#     # tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
-#     # model = AutoModelForCausalLM.from_pretrained(model_name)
-#     # 这里只是简单地返回 IDs 和损失值，但是可以根据实际需求添加颜色或其他样式
-#     return [1, 2], [0.1, 0.7]
-
-
-# def color_pipeline(text=["hi", "FreshEval"],  model=None):
-#     """
-#     给定一个文本，返回其对应的着色文本。
-#     """
-#     tokenizer=None
-#     ids, loss = get_ids_loss(text, tokenizer, model)
-#     text = get_text(ids, tokenizer)
-#     return color_text(text, loss)
-
-# # 创建 Gradio 界面
-# with gr.Blocks() as demo:
-#     with gr.Tab("color your text"):
-#         with gr.Row():
-#             text_input = gr.Textbox(label="input text", placeholder="input your text here...")
-#             # loss_input = gr.Number(label="loss")
-#             model_input = gr.Textbox(label="model name", placeholder="input your model name here...")
-#             color_text_output = gr.HTML(label="colored text")
-#             gr.Markdown("## Text Examples")
-#             # gr.Examples(
-#             #     [["hi", "Adam"], [0.1,0.7]],
-#             #     [text_input, loss_input],
-#             #     cache_examples=True,
-#             #     fn=color_text,
-#             #     outputs=color_text_output
-#             # )
-#         color_text_button = gr.Button("color the text").click(color_pipeline, inputs=[text_input, model_input], outputs=color_text_output)
-
-
-#         date_time_input = gr.Textbox(label="the date when the text is generated")#TODO add date time input
-#         description_input = gr.Textbox(label="description of the text")
-#         submit_button = gr.Button("submit a post or record")
-#         #TODO add model and its score
-
-    
-#     # with gr.Tab("ID 转文本展示"):
-#     #     with gr.Row():
-#     #         ids_input = gr.Textbox(label="输入 IDs (如 [101, 102, ...])")
-#     #         tokenizer_input = gr.Textbox(label="Tokenizer 名称", value="bert-base-uncased")
-#     #         show_text_output = gr.Textbox(label="转换后的文本")
-#     #     show_text_button = gr.Button("转换并展示").click(show_text, inputs=[ids_input, tokenizer_input], outputs=show_text_output)
-    
-#     with gr.Tab("model ppl with time"):
-#         '''
-#         see the matplotlib example, to see ppl with time, select the models
-#         '''
-
-    
-#     with gr.Tab("model ppl with time"):
-#         '''
-#         see the matplotlib example, to see ppl with time, select the models
-#         ''' 
-
+os.system('git clone https://github.com/EleutherAI/lm-evaluation-harness')
+os.system('cd lm-evaluation-harness')
+os.system('pip install -e .')
+# 第一个功能：基于输入文本和对应的损失值对文本进行着色展示
+def color_text(text_list=["hi", "FreshEval"], loss_list=[0.1,0.7]):
+    """
+    根据损失值为文本着色。
+    """
+    highlighted_text = []
+    for text, loss in zip(text_list, loss_list):
+        # color = "#FF0000" if float(loss) > 0.5 else "#00FF00"
+        color=loss
+        # highlighted_text.append({"text": text, "bg_color": color})
+        highlighted_text.append((text, color))
+
+        print(highlighted_text)
+    return highlighted_text
+
+# 第二个功能：根据 ID 列表和 tokenizer 将 ID 转换为文本，并展示
+def get_text(ids_list=[0.1,0.7], tokenizer=None):
+    """
+    给定一个 ID 列表和 tokenizer 名称，将这些 ID 转换成文本。
+    """
+    return ['Hi', 'Adam']
+    # tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
+    # text = tokenizer.decode(eval(ids_list), skip_special_tokens=True)
+    # 这里只是简单地返回文本，但是可以根据实际需求添加颜色或其他样式
+    # return text
+
+
+def get_ids_loss(text, tokenizer, model):
+    """
+    给定一个文本，model and its tokenizer,返回其对应的 IDs 和损失值。
+    """
+    # tokenizer = AutoTokenizer.from_pretrained(tokenizer_name)
+    # model = AutoModelForCausalLM.from_pretrained(model_name)
+    # 这里只是简单地返回 IDs 和损失值，但是可以根据实际需求添加颜色或其他样式
+    return [1, 2], [0.1, 0.7]
+
+
+def color_pipeline(text=["hi", "FreshEval"],  model=None):
+    """
+    给定一个文本，返回其对应的着色文本。
+    """
+    tokenizer=None # get tokenizer
+    ids, loss = get_ids_loss(text, tokenizer, model)
+    text = get_text(ids, tokenizer)
+    return color_text(text, loss)
+
+
+# TODO can this be global ? maybe need session to store info of the user
+
+# 创建 Gradio 界面
+with gr.Blocks() as demo:
+    with gr.Tab("color your text"):
+        with gr.Row():
+            text_input = gr.Textbox(label="input text", placeholder="input your text here...")
+            # TODO craw and drop the file
+
+            # loss_input = gr.Number(label="loss")
+            model_input = gr.Textbox(label="model name", placeholder="input your model name here...")
+            # TODO select models that can be used online
+            # TODO maybe add our own models
+
+
+            color_text_output = gr.HTML(label="colored text")
+            # gr.Markdown("## Text Examples")
+            # gr.Examples(
+            #     [["hi", "Adam"], [0.1,0.7]],
+            #     [text_input, loss_input],
+            #     cache_examples=True,
+            #     fn=color_text,
+            #     outputs=color_text_output
+            # )
+        color_text_button = gr.Button("color the text").click(color_pipeline, inputs=[text_input, model_input], outputs=gr.HighlightedText(label="colored text"))
+
+
+        date_time_input = gr.Textbox(label="the date when the text is generated")#TODO add date time input
+        description_input = gr.Textbox(label="description of the text")
+        submit_button = gr.Button("submit a post or record").click()
+        #TODO add model and its score
+
+    with gr.Tab('test your qeustion'):
+        '''
+        use extract, or use ppl
+        '''
+        question=gr.Textbox(placeholder='input your question here...')
+        answer=gr.Textbox(placeholder='input your answer here...')
+        other_choices=gr.Textbox(placeholder='input your other choices here...')
         
+        test_button=gr.Button('test').click()
+        #TODO add the model and its score
 
-# demo.launch()
+        def test_question(question, answer, other_choices):
+            '''
+            use extract, or use ppl
+            '''
+            answer_ppl, other_choices_ppl = get_ppl(question, answer, other_choices)
+            return answer_ppl, other_choices_ppl
 
 
 
-# import gradio as gr
-# from transformers import pipeline
+    with gr.Tab("model text ppl with time"):
+        '''
+        see the matplotlib example, to see ppl with time, select the models
+        '''
+        # load the json file with time,
 
+    
+    with gr.Tab("model quesion acc with time"):
+        '''
+        see the matplotlib example, to see ppl with time, select the models
+        ''' 
+        #
+
+
+    with gr.Tab("hot questions"):
+        '''
+        see the questions and answers
+        '''
+        with gr.Tab("ppl"):
+            '''
+            see the questions
+            '''
+        
 
-# pipeline = pipeline(task="image-classification", model="julien-c/hotdog-not-hotdog")
-
-# def predict(input_img):
-#     predictions = pipeline(input_img)
-#     return input_img, {p["label"]: p["score"] for p in predictions} 
+demo.launch(debug=True)
 
-# gradio_app = gr.Interface(
-#     predict,
-#     inputs=gr.Image(label="Select hot dog candidate", sources=['upload', 'webcam'], type="pil"),
-#     outputs=[gr.Image(label="Processed Image"), gr.Label(label="Result", num_top_classes=2)],
-#     title="Hot Dog? Or Not?",
-# )
 
-# if __name__ == "__main__":
-#     gradio_app.launch()
 
 
 
-import gradio as gr
+# import gradio as gr
+# import os
+# os.system('python -m spacy download en_core_web_sm')
+# import spacy
+# from spacy import displacy
+
+# nlp = spacy.load("en_core_web_sm")
+
+# def text_analysis(text):
+#     doc = nlp(text)
+#     html = displacy.render(doc, style="dep", page=True)
+#     html = (
+#         "<div style='max-width:100%; max-height:360px; overflow:auto'>"
+#         + html
+#         + "</div>"
+#     )
+#     pos_count = {
+#         "char_count": len(text),
+#         "token_count": 0,
+#     }
+#     pos_tokens = []
+
+#     for token in doc:
+#         pos_tokens.extend([(token.text, token.pos_), (" ", None)])
+
+#     return pos_tokens, pos_count, html
+
+# demo = gr.Interface(
+#     text_analysis,
+#     gr.Textbox(placeholder="Enter sentence here..."),
+#     ["highlight", "json", "html"],
+#     examples=[
+#         ["What a beautiful morning for a walk!"],
+#         ["It was the best of times, it was the worst of times."],
+#     ],
+# )
 
-def greet(name, intensity):
-    return "Hello, " + name + "!" * int(intensity)
+# demo.launch()
 
-demo = gr.Interface(
-    fn=greet,
-    inputs=["text", "slider"],
-    outputs=["text"],
-)
 
-demo.launch(debug=True)
 
-# lm-eval 
-# lm-evaluation-harness 
+# # lm-eval 
+# # lm-evaluation-harness 

diff_color.py

@@ -0,0 +1,42 @@
+
+from difflib import Differ
+
+import gradio as gr
+
+
+def diff_texts(text1, text2):
+    d = Differ()
+    rtn =[
+        (token[2:], token[0] if token[0] != " " else None)
+        for token in d.compare(text1, text2)
+    ]
+    print(rtn)
+    return rtn
+
+
+demo = gr.Interface(
+    diff_texts,
+    [
+        gr.Textbox(
+            label="Text 1",
+            info="Initial text",
+            lines=3,
+            value="The quick brown fox jumped over the lazy dogs.",
+        ),
+        gr.Textbox(
+            label="Text 2",
+            info="Text to compare",
+            lines=3,
+            value="The fast brown fox jumps over lazy dogs.",
+        ),
+    ],
+    gr.HighlightedText(
+        label="Diff",
+        combine_adjacent=True,
+        show_legend=True,
+        color_map={"+": "red", "-": "green"}),
+    theme=gr.themes.Base()# the return is here
+)
+if __name__ == "__main__":
+    demo.launch()
+

flagged/log.csv

@@ -0,0 +1,2 @@
+Text 1,Text 2,Diff,flag,username,timestamp
+The quick brown fox jumped over the lazy dogs.,The fast brown fox jumps over lazy dogs.,,,,2024-03-13 16:50:01.853095

hello_test.py

@@ -0,0 +1,13 @@
+import gradio as gr
+
+def greet(name, intensity):
+    return "Hello, " + name + ",,!" * int(intensity)
+    # return "Hello, " + name + ",,!" * int(0/int(intensity))# you can see the bug in command line
+
+demo = gr.Interface(
+    fn=greet,
+    inputs=["text", "slider"],
+    outputs=["text"],
+)
+
+demo.launch(debug=True)

hf_space_test.py

@@ -0,0 +1,20 @@
+# this need hugginface connection
+import gradio as gr
+from transformers  import pipeline
+
+
+pipeline = pipeline(task="image-classification", model="julien-c/hotdog-not-hotdog")
+
+def predict(input_img):
+    predictions = pipeline(input_img)
+    return input_img, {p["label"]: p["score"] for p in predictions} 
+
+gradio_app = gr.Interface(
+    predict,
+    inputs=gr.Image(label="Select hot dog candidate", sources=['upload', 'webcam'], type="pil"),
+    outputs=[gr.Image(label="Processed Image"), gr.Label(label="Result", num_top_classes=2)],
+    title="Hot Dog? Or Not?",
+)
+
+if __name__ == "__main__":
+    gradio_app.launch(debug=True)

lm-evaluation-harness

@@ -0,0 +1 @@
+Subproject commit 49695e8d94c3ab011b7ae8814d809de30b1b1182